%0 Conference Proceedings
%T BeamAggR: Beam Aggregation Reasoning over Multi-source Knowledge for Multi-hop Question Answering
%A Chu, Zheng
%A Chen, Jingchang
%A Chen, Qianglong
%A Wang, Haotian
%A Zhu, Kun
%A Du, Xiyuan
%A Yu, Weijiang
%A Liu, Ming
%A Qin, Bing
%Y Ku, Lun-Wei
%Y Martins, Andre
%Y Srikumar, Vivek
%S Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)
%D 2024
%8 August
%I Association for Computational Linguistics
%C Bangkok, Thailand
%F chu-etal-2024-beamaggr
%X Large language models (LLMs) have demonstrated strong reasoning capabilities.Nevertheless, they still suffer from factual errors when tackling knowledge-intensive tasks.Retrieval-augmented reasoning represents a promising approach.However, significant challenges still persist, including inaccurate and insufficient retrieval for complex questions, as well as difficulty in integrating multi-source knowledge.To address this, we propose Beam Aggregation Reasoning (BeamAggR), a reasoning framework for knowledge-intensive multi-hop QA.BeamAggR explores and prioritizes promising answers at each hop of question.Concretely, we parse the complex questions into trees, which include atom and composite questions, followed by bottom-up reasoning.For atomic questions, the LLM conducts reasoning on multi-source knowledge to get answer candidates.For composite questions, the LLM combines beam candidates, explores multiple reasoning paths through probabilistic aggregation, and prioritizes the most promising trajectory.Extensive experiments on four open-domain multi-hop reasoning datasets show that our method significantly outperforms SOTA methods by 8.5%.Furthermore, our analysis reveals that BeamAggR elicits better knowledge collaboration and answer aggregation.
%R 10.18653/v1/2024.acl-long.67
%U https://aclanthology.org/2024.acl-long.67
%U https://doi.org/10.18653/v1/2024.acl-long.67
%P 1229-1248